NONMEM Users Network Archive

Hosted by Cognigen

Format for submission to FDA

From: peter.l.bonate
Date: Tue, 30 Jun 2009 08:44:28 -0400

I've had this request many times from the FDA (NONMEM datasets as SAS
transport files) and have complied each time. Although NONMEM does not
read XPT files, they are easy to convert to TXT files. The nice thing
about XPT files is that the labels can accomany the columns. Keep in mind =

though that even though the current versions of SAS can have long label
names (20 characters I think) and long label descriptors (up to 80
characters), the FDA wants the old, archaic variable names and labels.
Variable names of no more than 8 characters and labels of less than 40
characters (this last I one I am unsure of the exact number).

Peter L. Bonate, PhD, FCP, FAAPS
Director, Clinical Pharmacology, Modeling, and Simulation
5 Moore Drive, 17.2259
Research Triangle Park, NC 27709
email: peter.l.bonate
phone: 919-483-7534
fax: 919-483-8948

?There is nothing wrong with change, if it is in the right direction? -
Winston Churchill

"MIDO (Michael Dodds)" <doddsm
Sent by: owner-nmusers
29-Jun-2009 18:20

RE: [NMusers] Format for submission to FDA


I would also be very interested in your responses. Recently, we've
attempted to submit a Berkeley Madonna file as part of a submission, and
that file type has caused some problems. While it is possible to take the =

BM code out as a flat .txt and the embedded data set out as a .csv or .xls =

(or even .xpt, I suppose), I questioned the utility of that.
"Reconstituting" the analysis from the components might be possible here,
but it's error prone and, frankly, inefficient.

In the end, we were able to simple embed the BM file (extension .mmd) into =

the eCTD format and send it along. Someone without the BM tool installed
on their computer would receive a puzzled response from their web browser
when trying to open the file. However, someone with the tool installed is =

able to correctly open the BM file.

Again, comments from our regulatory colleagues would be very helpful here. =

 How can we package and transport complex analysis (my example was single
BM file, Dennis suggests a series of NONMEM runs in a file structure) and
share that with regulatory bodies?

Mike Dodds

-----Original Message-----
From: owner-nmusers
On Behalf Of Dennis Fisher
Sent: Monday, June 29, 2009 1:36 PM
To: nmusers
Subject: [NMusers] Format for submission to FDA


One of my clients is approaching submission of an NDA to FDA. I did
the pop PK analysis; the files that I generated are in three formats -
text, PDF, and csv. NONMEM text output is saved in two formats (text
and PDF); NONMEM tables are saved in two formats (text and csv);
graphics created in R are saved in PDF format. Datasets used for each
step of the analysis are archived in two formats, text (an exact copy
of the input file plus an exact copy of the FDATA file) and csv (the
FDATA file formatted as csv).

Our plan (based on my other recent submissions to FDA, none of which
has elicited any complaints from them), was to submit these documents
in the formats described above. However, my client received a request
from FDA that:
                 "datasets [used in the analysis] should be submitted as
SAS transport
(*.xpt) files".
In addition, the company managing the submission wants to convert all
my other csv files to xpt format (although there was no specific
request for this).

Can anyone (particularly anyone at FDA) clarify this for me? In that
NONMEM will not accept .xpt files, I am surprised that FDA would want
datasets in that format. Should my client honor the request or should
they explain to the project manager that NONMEM requires text files so
we plan to submit these files in text format (either .txt or csv).

Thanks for any insights on this.


Dennis Fisher MD
P < (The "P Less Than" Company)
Phone: 1-866-PLessThan (1-866-753-7784)
Fax: 1-415-564-2220

Received on Tue Jun 30 2009 - 08:44:28 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: